This implementation is based on the reinforcement learning algorithm implementations by OpenAI Baselines (https://github.com/openai/baselines) which is open-source.

Prerequisites
python3 (>=3.5), system packages CMake, OpenMPI and zlib.
Ubuntu:
sudo apt-get update && sudo apt-get install cmake libopenmpi-dev python3-dev zlib1g-dev
Mac OS X:
brew install cmake openmpi

Tensorflow versions 1.4-1.14

Installation
Clone the repo and cd into it:
git clone https://github.com/openai/baselines.git
cd baselines
pip install tensorflow==1.14
pip install -e .

Train with the following command
OPENAI_LOG_FORMAT=tensorboard,stdout,csv,log python -m baselines.run_exp --env=VentureNoFrameskip-v4 --num_timesteps=1e10 --save_path=./models/Venture_combo/Venture_combo_0-$seed --log_path=./logs/Venture_combo/Venture_combo_0-$seed --num_env=4 --seed=$seed --load_path=0

--env can be used to specify different game environment